摘要： 提出了一种针对序列定标图像的三维空间点精确欧氏重建算法，相对于传统基于代数误差最小化的方法，该算法用几何误差作为优化的目标函数，从而保证了三维重建点在欧氏距离误差的意义下是最优的；该算法是线性的，计算效率高；此外还可以扩展到任意多幅图像的重建中。真实测试图像的实验结果以及与传统算法的比较验证了该算法的有效性和精确性。

关键词: 三维重建, 三角形法, 代数误差, 几何误差

Abstract: This paper proposes an accurate linear triangulation algorithm based on image sequence. Compared with traditional algebraic error based method, this algorithm uses geometrical error as the cost function, which ensures that the results are the best under physical meanings of Euclidean distance. This method is also linear, which is time efficient. In addition, the method can be extended to any number of images. Experimental results of real images and comparisons with traditional method demonstrate its feasibility and accuracy.

Key words: 3D reconstruction, Triangulation, Algebraic error, Geometrical error

中图分类号: 

• TP301